As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Start by deciding which of three things has failed: the display, the image, or the number. A blank or frozen screen is an electronics or power fault. A soft or noisy image is usually focus, a dirty window, or a resolution limit. A temperature that looks wrong is usually emissivity, reflected background, or a reading taken outside the camera’s stated range. Check the simple causes first: lens cap, dirty germanium window, battery, and whether the camera is in the right measurement mode. Only after those should you suspect the detector.
The screen is blank, frozen or will not power on
If the display shows nothing, the fault is in the camera as an electronic device, not in the thermal physics. Work through power first. Most handheld thermal cameras run on removable cells or internal lithium packs, and a pack that has been stored flat for months may take a charge but not hold enough to boot the detector. Try a known-good cell or a mains supply if the listing supports one.
If the camera powers on but the screen stays black, look at the lens. A germanium window looks opaque to the eye, so a lens cap left on gives a completely uniform image that can read as a dead screen. Some cameras also ship with a protective film over the window that must be removed. A window smeared with fingerprints scatters long-wave infrared and produces a soft, low-contrast image that owners often describe as the camera having failed.
If the screen is frozen with a live-looking image, the processor has locked up. Remove the battery, wait, and restart. A camera that freezes repeatedly during image saving is usually writing to a full or slow memory card. The DCTOPT listing, for example, specifies a 4GB detachable card and storage for over 10,000 images, so a card that has been swapped for a smaller or slower one is a plausible cause of save-time lockups.
One more electronic cause deserves mention because it is invisible: the shutter. Uncooled microbolometer cameras recalibrate by passing a flag across the detector, which produces a brief click and a uniform image. If the shutter sticks, the camera may show a flat grey field or a persistent ghost pattern. That is a service fault, not something the user can fix.
The image is soft, grainy or shows no detail
A thermal image that looks like a smeared watercolour is the most common complaint, and it is almost never a failed detector. Three causes account for most cases: focus, resolution and the scene itself.
Focus first. Fixed-lens cameras are focused at a set distance, often a metre or two, and anything much closer or further is soft. The MiLESEEY TR20 uses a 50 degree wide-angle lens, which covers a lot of wall in one frame but spreads its 320 by 240 detector elements over a wide field, so each pixel covers a larger patch of surface than a narrow lens would. A wide lens is the right choice for scanning a room and the wrong choice for reading a small component on a board. If your subject is small, move closer and accept that the image will soften, or use a camera with a narrower field.
Resolution is the second cause. A 192 by 192 sensor upscaled to 384 by 384, as the AMPBANK M192 listing describes, produces a smoother-looking picture but does not create detail that the detector did not capture. The distinction matters: super-resolution interpolation makes an image easier to read, but the true optical resolution still sets the smallest object you can measure. The explanation of thermal resolution covers how to calculate the spot size a camera can resolve at a given distance.
The third cause is the scene. A wall at uniform temperature has no thermal contrast to show, so the camera displays noise. Point it at a surface with a real temperature difference, such as a radiator pipe against plaster, and detail appears. Graininess in a uniform scene is normal and is quantified by noise equivalent temperature difference, or NETD. The TR20 publishes NETD below 40 millikelvin and the DCTOPT publishes below 50 millikelvin, both of which are sensitive enough to show small differences once the scene actually contains them. A camera with a poor NETD shows grain even on a good scene; a camera with a good NETD shows grain only where there is nothing to see.
The temperature reading looks wrong
This is where most owners conclude the camera is faulty, and where the camera is most often right and the technique is wrong. A thermal camera does not measure temperature directly. It measures the infrared radiance leaving a surface and solves for temperature using an assumed emissivity and an assumed reflected background temperature. Get either assumption wrong and the number is wrong, even though the detector is working perfectly.
Emissivity is the fraction of ideal black-body radiation a surface emits. Bare shiny metal has low emissivity, often below 0.2, which means it emits little of its own radiation and reflects most of what arrives from elsewhere. The camera reads the reflection and reports a temperature that can be far from the metal’s true temperature. The guide to shiny metal and thermal cameras explains the mechanism and the fixes: apply a matt tape or paint patch of known emissivity, or set the camera’s emissivity value to match the surface. The AMPBANK M192 listing states that emissivity is adjustable in the app, which is the control you need for this.
Reflected background is the second trap. A cold window or a warm radiator behind the camera is reflected by a low-emissivity surface and added to the apparent temperature. Most cameras let you set the reflected temperature, and the AMPBANK listing mentions ambient temperature settings in the app. If your camera does not expose that control, avoid measuring shiny surfaces where a strong source is reflected.
Range is the third trap. Every camera has a stated measurement range. The TR20 publishes -4 to 1022 degrees Fahrenheit, the AMPBANK publishes -4 to 842 degrees Fahrenheit, and the DCTOPT does not publish a range in the listing bullets. Outside the range the camera either clips, showing a saturated colour with no meaningful number, or refuses to display a value. A reading taken at the edge of the range is not trustworthy even when it appears.
Finally, check the spot size. If the target is smaller than the camera’s resolution element at that distance, the reading is an average of the target and its surroundings. The note on NETD explains why sensitivity and spot size are separate limits.
When the camera is working and the job is still wrong
Some faults are not faults. A thermal camera sees surface temperature, and it sees it through a lens that cannot pass visible light. That makes it the wrong instrument for several jobs that owners attempt anyway.
It cannot see through a wall. It shows the temperature of the wall surface, which may be influenced by what is behind it, but it does not image the cavity. Finding a pipe behind plaster relies on the pipe warming or cooling the plaster enough to create a pattern, and on the pattern being larger than the camera’s resolution element. A camera with a wide lens and a modest detector may not resolve a narrow pipe at room distance. A borescope is the instrument for seeing inside a cavity; the thermal camera is the instrument for finding where to look.
It cannot measure air temperature. It measures surfaces. A camera pointed at a room reports the temperature of the walls and furniture, not the air, and the two can differ by several degrees. For air temperature a thermometer is the right tool, which is why the comparison with an infrared thermometer matters: the spot thermometer and the camera share the same emissivity problem, but the camera shows the pattern and the thermometer gives one number.
It cannot see through glass. Ordinary window glass is opaque to long-wave infrared, so a camera pointed through a window shows the glass surface temperature, not the scene beyond. Germanium windows are used precisely because they pass the band; a glass window will not.
And it cannot replace a contact thermometer for absolute accuracy. The TR20 publishes plus or minus 2 percent measurement accuracy, and the AMPBANK publishes a maximum of plus or minus 2 degrees Celsius or 2 percent. Those are typical figures for the class, and they mean a reading of 100 degrees could be off by 2 degrees before any emissivity error is added. For a compliance measurement, use a contact probe and use the camera to find where to place it.
A fault-finding order that saves time
Work through these checks in order. Each one eliminates a class of cause before you move to the next.
| Symptom | First check | Second check | If both pass |
|---|---|---|---|
| No display | Battery or power supply | Lens cap and protective film | Service fault |
| Frozen image | Restart with battery removed | Memory card free space and speed | Service fault |
| Soft image | Focus distance | Window cleanliness | Resolution limit for the subject |
| Grainy image | Scene thermal contrast | NETD specification | Normal behaviour |
| Reading too high or low | Emissivity setting | Reflected background setting | Outside stated range or spot size too large |
| No reading at all | Measurement mode enabled | Target within range | Spot size smaller than resolution element |
The table is deliberately ordered from the cheapest check to the most expensive. Most owners who believe their camera has failed find the cause in the first two rows.
One product note makes the point concrete. The Bosch PRO GTC600C listing emphasises image storage, voice notes and a simple interface, and publishes no detector resolution, NETD or accuracy figure in its bullets. That does not make it a bad camera, but it means a buyer cannot diagnose a soft image against a published resolution limit. Where a listing is silent on a figure, the honest position is that the figure is not stated in the listing bullets, and the fault-finding has to rely on the checks above rather than on a specification comparison.
The camera is usually right and the setup is usually wrong
A thermal camera that appears to have failed is more often pointed at the wrong surface, set to the wrong emissivity, or asked to resolve something smaller than its detector can see. Check power, lens and focus first, then check the physics of the reading. If those pass and the image is still wrong, the camera needs service. If they pass and the reading is still wrong, the job needs a different instrument.
Frequently Asked Questions
Why does my thermal camera show a blank screen when it powers on?
The most common causes are a lens cap left on, a protective film still over the germanium window, or a battery that boots the electronics but not the detector. Remove the cap and film, then try a known-good battery. If the screen stays uniform after that, the detector or shutter is likely at fault and the camera needs service.
Why is my thermal image blurry even when the camera is new?
Fixed-lens thermal cameras focus at a set distance, so subjects much closer or further are soft. A wide-angle lens spreads a modest detector over a large field, which limits small-object detail. Clean the window, move to the focus distance, and check whether your subject is larger than the camera's resolution element at that range.
Why does my thermal camera read the wrong temperature on metal?
Bare shiny metal has low emissivity, so it emits little radiation of its own and reflects its surroundings. The camera solves for temperature using an assumed emissivity, and the assumption is wrong for the metal. Apply matt tape or paint of known emissivity, or set the camera's emissivity to match the surface, and set the reflected background temperature.
Can a thermal camera see through a wall or through glass?
No. It images surface temperature. A wall shows its own surface temperature, which may be influenced by what is behind it, but the cavity is not imaged. Ordinary glass is opaque to long-wave infrared, so a camera pointed through a window reads the glass, not the scene beyond.
My camera reads a temperature outside its stated range. Is it broken?
Probably not. Outside the stated range the camera clips or refuses to display a value, and a reading at the edge of the range is not trustworthy. Check the published range for your model. The MiLESEEY TR20 publishes -4 to 1022 degrees Fahrenheit and the AMPBANK M192 publishes -4 to 842 degrees Fahrenheit. If your listing does not publish a range, the figure is not stated in the listing bullets.
Why does the image look grainy on a plain wall?
A uniform surface has no thermal contrast, so the camera displays its own noise. Graininess in a featureless scene is normal and is described by the NETD figure. Point the camera at a surface with a real temperature difference and detail should appear. If grain persists on a good scene, the NETD may be poor or the detector may need recalibration.